How to get cited by ChatGPT

Ask ChatGPT which tool to buy, which agency to hire, or how to solve a problem in your category, and it will often answer with a short list of named brands. In its search mode it goes further and links the sources it used. Getting cited by ChatGPT means being one of those brands it names, or one of those pages it links, when it answers the questions your buyers actually ask.

This is a practical playbook. One caveat up front: ChatGPT’s exact citation behavior is not fully documented, and it varies by prompt, by account and over time. So the steps below are not tricks that force a mention. They are the durable signals that, together, make you the kind of source ChatGPT reaches for.

How ChatGPT surfaces and cites sources

You cannot influence what you do not understand, so start with the mechanics. ChatGPT surfaces brands in two broad ways, and getting cited means addressing both. This is also what people mean when they ask how to rank on ChatGPT: it is less a ranked list than a question of whether you are reachable, relevant and trusted enough to be pulled into the answer.

Live search: it retrieves and links real pages

When ChatGPT browses the web to answer, it runs a search behind the scenes, reads the top results, and synthesizes a reply with links to its sources. ChatGPT’s search has drawn on Bing’s index, and OpenAI has also built its own crawling and retrieval on top, so the exact mix of what powers any given answer is not something OpenAI fully discloses. What matters for you is simpler: to be pulled into a browsed answer, your page has to be crawlable, clearly relevant to the question, and easy to lift a clean answer from. That is the same core work as answer engine optimization.

Training memory: it recalls how the web describes you

When ChatGPT answers from memory without browsing, it is drawing on patterns in the data it was trained on. There it can only mention brands that are well represented across the web it learned from: your own site, but also Reddit, Wikipedia, industry publications, review sites and comparison lists. A brand described the same way in many trustworthy places becomes a recognizable entity the model can recall and recommend. A brand that barely appears, or is described inconsistently, does not. Training data also has a cutoff, so a brand-new site will not exist in memory until a later model learns about it.

The 8-step playbook to get cited by ChatGPT

Here is the sequence we recommend, ordered by impact. The early steps are the foundation, the later ones compound, and none of them is a payment or a trick.

1. Let the AI crawlers in

None of the rest matters if OpenAI’s bots cannot read your pages. Check that your robots.txt allows the crawlers you want: GPTBot and OAI-SearchBot for OpenAI, plus ClaudeBot, PerplexityBot and Google-Extended if you want the other engines too. Just as important, make sure your real content is server-rendered into the HTML, not assembled only after JavaScript runs in the browser, because crawlers often will not execute it. A surprising share of “we are invisible in AI” cases come down to a blocked bot or content that only exists client-side.

2. Answer the exact question in the first two sentences

Answer engines reward passages they can extract without stitching them together. Open each section with a direct, self-contained answer to the question it is about, then explain underneath. If a buyer would ask “what is the best tool for small teams,” the first sentence or two should name the answer plainly before you add nuance. Lead with the conclusion, support it after. This is the single most controllable lever you have.

3. Shape your headings as real buyer questions

ChatGPT is matching a user’s question to content that visibly answers it, so phrase your headings the way buyers actually ask, in full questions rather than one-word labels. “How much does it cost?” beats “Pricing.” Cover the comparison and decision questions, not just the definitional ones, because those are the prompts where a brand tends to get named.

4. Add FAQ and honest comparison content

Two formats punch above their weight for citations. A genuine FAQ section answers the long tail of follow-up questions in exactly the tight, liftable form engines favor. Honest comparison content, how you stack up against alternatives and what you are and are not the best fit for, is what ChatGPT leans on when a buyer asks it to compare options. Write both for the reader first: thin content stuffed with question headings fools no one.

5. Strengthen your brand entity

Models recommend entities they recognize. Describe your company the same way everywhere it appears, publish Organization schema with your real identifiers and profiles, and keep the basic facts about who you are and what you do consistent across every source. Being referenced on trusted, independent sites, and where it is genuinely warranted a Wikidata or Wikipedia presence, turns a name into an entity the model can place. You can also publish an llms.txt file as a low-cost map of your best content, though be honest that it is an emerging signal, not a citation lever: see our honest guide to llms.txt.

6. Earn mentions on the sites ChatGPT trusts

This is the hardest step and the one that most separates brands that get cited from brands that do not. ChatGPT leans on sources it already treats as credible, so you want to be present where it looks: reputable press, active and genuinely helpful participation in communities like Reddit, and the review and “best tools” lists people reach for when they compare options. You cannot buy your way onto these in a way that lasts. You earn the mentions by being worth mentioning.

7. Publish original data worth quoting

A single piece of original research, a benchmark, a survey of your users, or a number no one else has, does more for citability than a dozen restated explainers. Models, and the writers who feed them, gravitate to concrete, quotable claims with a clear owner. If a statistic becomes “the” number people reference in your category, you become the source attached to it.

8. Measure which prompts cite you, then close the gaps

Getting cited by ChatGPT is a loop, not a launch. Ask the engine your buyer questions on a schedule, record who it names and links, and work the specific prompts where a competitor shows up and you do not. The next section covers how to run that check yourself.

The outsized role of third-party mentions

Step 6 is the one most brands underrate, so it is worth showing why with the data. The most consistent finding in recent citation research is how little of what ChatGPT cites comes from brands’ own websites, and how much comes from third-party sources: editorial coverage, review platforms, community threads and reference sites.

A September 2025 study from four University of Toronto researchers, measuring ChatGPT, Perplexity and Gemini, found that AI search shows “a systematic and overwhelming bias towards Earned media”, third-party, authoritative sources, over brand-owned and social content. In their U.S. data, earned media accounted for roughly 82% of AI citations in the automotive category and about 92% in consumer electronics, where Google, by contrast, drew a far more balanced mix that included much more brand-owned content. Independently, Profound’s analysis of about 730,000 real ChatGPT conversations from late 2025 found the citation pool is strikingly fragmented and third-party: the ten most-cited domains together accounted for only about 12% of all citations, no single source dominated, and the most-cited was Wikipedia, appearing in nearly one in six conversations that carried citations, followed by Reddit.

This is why the “best [category] tools” listicles, the review platforms your buyers check, and the community threads where your category is discussed matter so much: those are precisely the earned, third-party sources ChatGPT leans on, especially when it answers from memory without browsing. Being named in the credible lists and profiles people already trust does more for your citability than another explainer on your own blog. You cannot reliably buy your way onto them, and thin or planted mentions do not last, but genuine, earned presence on the sources ChatGPT already reads is among the most durable levers you have.

How to check if ChatGPT cites your brand

You can start measuring by hand today. Make a list of the questions a buyer in your category would actually type, the commercial ones especially: “best tool for X,” “alternatives to Y,” “is Z worth it.” Ask ChatGPT each one, with search on and off, and record whether it names you, links you, or leaves you out. Note who it does cite instead, because that is your competitive gap in plain sight.

Do it more than once. ChatGPT’s answers vary between sessions and accounts, so a single check is noise, not signal. Ask the same questions on a schedule, in fresh chats, and track how your citation rate and your share of the named brands move over time. Pair that with the click side by connecting Google Search Console, so you can watch real impressions and clicks respond as your fixes land rather than trusting a vanity score.

This is exactly what Cituna automates: it runs the prompts, tracks citations across ChatGPT, Perplexity, Gemini, Claude, Grok and Google AI Overviews over time, flags the gaps and drafts the fixes, and it exposes a built-in MCP server so you can query your own scans and gaps from inside Claude. However you do it, the principle holds: watch the answers, then work the gaps. For the wider measurement picture see AI visibility, and for the sibling playbooks see how to rank on Perplexity and Google AI Overviews.

Measure which prompts cite you

The section above shows how to check by hand; measuring turns those checks into numbers you can move. Watch three: your citation rate (the share of your buyer prompts where ChatGPT names or links you), your share of the named brands (you versus each competitor in the same answers), and your prominence (whether you are named first or buried mid-answer). Track them in fresh chats, with search on and off, because a single run is noise.

Doing that daily by hand is tedious, which is the case for a ChatGPT rank tracker: it runs your prompts on a schedule and records the citation, the position when you are cited, and the competitors named in your place, as a trend rather than a one-off. Cituna runs the same daily matrix across the six engines, ChatGPT, Perplexity, Gemini, Claude, Grok and Google AI Overviews (not Microsoft Copilot), pairs it with Google Search Console so you can tie answer-side gains to real clicks, and turns each gap into a concrete fix. The principle stays the same whether you automate it or not: watch the answers, then work the gaps.

Common mistakes

• Blocking the crawlers you want. An overzealous firewall rule or a blanket bot block can quietly exclude GPTBot or OAI-SearchBot. Check before you assume you are being ignored.
• Hiding content behind JavaScript. If your answer only appears after scripts run, a crawler may never see it. Server-render what you want cited.
• Writing for the model instead of the reader. ChatGPT is trying to surface genuinely useful content, so question-shaped headings over thin filler fool no one and can read as spam.
• Treating schema as a growth hack. Structured data helps machines parse you, but on its own it does not manufacture trust or force a citation.
• Expecting to pay your way in. There is no ad slot or paid placement inside ChatGPT’s organic answers. You earn the mention or you do not.
• Checking once and giving up. Answers vary run to run, so a single query tells you little. Measure on a schedule and work the gaps you find.

Do the fundamentals well, earn a little trust on the sources ChatGPT already reads, and measure the loop, and you move from invisible to cited. That is the whole game.